Photography

The emergence of generative artificial intelligence (GenAI) technology, specifically text-to-image and image-to-image diffusion models capable of generating photorealistic images, presents a significant disruption to photographic practitioners, businesses, and viewers. Whilst AI-generated images are not technically photographs, they can be indistinguishable from photographs, with the same capacity to represent reality and evoke emotional responses. This chapter identifies how GenAI is disrupting the field of photography and explores potential consequences of this disruption. The most significant disruption in the history of photography was the transition from analog to digital that accelerated in the final decade of the twentieth century. An outline of the impacts of this shift on photographers, photographic businesses, and consumers of photographic images provides background for the foregoing discussion focussed on three crucial concerns around GenAI’s impact on photographic industries: ethical transparency and diversification of traditional business models; mitigating bias and homogenization in photorealistic GenAI imagery; and innovation in artistic expression catalyzed by the adoption of Generative AI in the visual arts. Ethical concerns and legal challenges around use of copyright material by companies to train GenAI models are summarized. Companies, including OpenAI, that admit to unauthorized use of copyrighted material for GenAI training are contrasted with those, including Adobe, that claim to offer ethical GenAI products trained on datasets over which they hold copyright. It is argued that artists, whose work is at risk of being used to train GenAI models, and those seeking indemnity from copyright infringement claims in their commercial use of GenAI imagery, are more likely to subscribe to businesses that use ethical training sets and indemnify the use of images generated on their platforms. The fact that GenAI models have enabled companies taking ethical stances to swiftly diversify their product offerings considerably is discussed. It is argued that when image editing software can create images from text prompts and businesses that offer stock photographs can incorporate image editing capabilities, the risk of market concentration grows significantly. Strategies to mitigate bias in, and homogenization of, GenAI imagery, as well as the risk of devaluation of photographic imagery, are also identified. Finally, three artist projects that incorporate AI-generated imagery exemplify how GenAI has enabled new avenues for artistic expression. Sara Oscar’s 2023-ongoing project, Counterfactual Departures, sees the artist using GenAI to create photographic images of her Thai mother’s migration to Australia in 1974, an event of which there is no photographic record. The project invites discussion about the role of photography in memory-making and how AI-generated imagery is connected to traditional forms of photography. Julia Scott-Stevenson’s Collective Visions (2024) uses GenAI as a tool to enable participants to envision their hopes and possible fears of climate futures in an immersive 360° environment. Justin Harvey’s 2023 work, Unprompted Studies 1–3, is a triptych of animated image sequences representing three studies in machine-generated animation devoid of text input, for which diffusion models were instructed to create images “unprompted.” The work elicits questions around well-noted biases evident in early diffusion models (white, western, male, etc.) and how artists might assemble unprompted outcomes of generative AI models to make new meaning.